Btrfs: move data checksumming into a dedicated tree

Btrfs stores checksums for each data block.  Until now, they have
been stored in the subvolume trees, indexed by the inode that is
referencing the data block.  This means that when we read the inode,
we've probably read in at least some checksums as well.

But, this has a few problems:

* The checksums are indexed by logical offset in the file.  When
compression is on, this means we have to do the expensive checksumming
on the uncompressed data.  It would be faster if we could checksum
the compressed data instead.

* If we implement encryption, we'll be checksumming the plain text and
storing that on disk.  This is significantly less secure.

* For either compression or encryption, we have to get the plain text
back before we can verify the checksum as correct.  This makes the raid
layer balancing and extent moving much more expensive.

* It makes the front end caching code more complex, as we have touch
the subvolume and inodes as we cache extents.

* There is potentitally one copy of the checksum in each subvolume
referencing an extent.

The solution used here is to store the extent checksums in a dedicated
tree.  This allows us to index the checksums by phyiscal extent
start and length.  It means:

* The checksum is against the data stored on disk, after any compression
or encryption is done.

* The checksum is stored in a central location, and can be verified without
following back references, or reading inodes.

This makes compression significantly faster by reducing the amount of
data that needs to be checksummed.  It will also allow much faster
raid management code in general.

The checksums are indexed by a key with a fixed objectid (a magic value
in ctree.h) and offset set to the starting byte of the extent.  This
allows us to copy the checksum items into the fsync log tree directly (or
any other tree), without having to invent a second format for them.

Signed-off-by: Chris Mason <chris.mason@oracle.com>

1 files changed, 69 insertions, 116 deletions

diff --git a/fs/btrfs/file-item.c b/fs/btrfs/file-item.c
index 234ed441736c..a3ad2ce00116 100644
--- a/fs/btrfs/file-item.c
+++ b/fs/btrfs/file-item.c
@@ -74,8 +74,7 @@ out:
 struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans,
 					  struct btrfs_root *root,
 					  struct btrfs_path *path,
-					  u64 objectid, u64 offset,
-					  int cow)
+					  u64 bytenr, int cow)
 {
 	int ret;
 	struct btrfs_key file_key;
@@ -87,9 +86,9 @@ struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans,
 		btrfs_super_csum_size(&root->fs_info->super_copy);
 	int csums_in_item;
 
-	file_key.objectid = objectid;
-	file_key.offset = offset;
-	btrfs_set_key_type(&file_key, BTRFS_CSUM_ITEM_KEY);
+	file_key.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
+	file_key.offset = bytenr;
+	btrfs_set_key_type(&file_key, BTRFS_EXTENT_CSUM_KEY);
 	ret = btrfs_search_slot(trans, root, &file_key, path, 0, cow);
 	if (ret < 0)
 		goto fail;
@@ -100,11 +99,10 @@ struct btrfs_csum_item *btrfs_lookup_csum(struct btrfs_trans_handle *trans,
 			goto fail;
 		path->slots[0]--;
 		btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
-		if (btrfs_key_type(&found_key) != BTRFS_CSUM_ITEM_KEY ||
-		    found_key.objectid != objectid) {
+		if (btrfs_key_type(&found_key) != BTRFS_EXTENT_CSUM_KEY)
 			goto fail;
-		}
-		csum_offset = (offset - found_key.offset) >>
+
+		csum_offset = (bytenr - found_key.offset) >>
 				root->fs_info->sb->s_blocksize_bits;
 		csums_in_item = btrfs_item_size_nr(leaf, path->slots[0]);
 		csums_in_item /= csum_size;
@@ -143,7 +141,7 @@ int btrfs_lookup_file_extent(struct btrfs_trans_handle *trans,
 }
 
 int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
-			  struct bio *bio)
+			  struct bio *bio, u32 *dst)
 {
 	u32 sum;
 	struct bio_vec *bvec = bio->bi_io_vec;
@@ -151,6 +149,7 @@ int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
 	u64 offset;
 	u64 item_start_offset = 0;
 	u64 item_last_offset = 0;
+	u64 disk_bytenr;
 	u32 diff;
 	u16 csum_size =
 		btrfs_super_csum_size(&root->fs_info->super_copy);
@@ -165,21 +164,22 @@ int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
 
 	WARN_ON(bio->bi_vcnt <= 0);
 
+	disk_bytenr = (u64)bio->bi_sector << 9;
 	while(bio_index < bio->bi_vcnt) {
 		offset = page_offset(bvec->bv_page) + bvec->bv_offset;
-		ret = btrfs_find_ordered_sum(inode, offset, &sum);
+		ret = btrfs_find_ordered_sum(inode, offset, disk_bytenr, &sum);
 		if (ret == 0)
 			goto found;
 
-		if (!item || offset < item_start_offset ||
-		    offset >= item_last_offset) {
+		if (!item || disk_bytenr < item_start_offset ||
+		    disk_bytenr >= item_last_offset) {
 			struct btrfs_key found_key;
 			u32 item_size;
 
 			if (item)
 				btrfs_release_path(root, path);
-			item = btrfs_lookup_csum(NULL, root, path,
-						 inode->i_ino, offset, 0);
+			item = btrfs_lookup_csum(NULL, root->fs_info->csum_root,
+						 path, disk_bytenr, 0);
 			if (IS_ERR(item)) {
 				ret = PTR_ERR(item);
 				if (ret == -ENOENT || ret == -EFBIG)
@@ -208,7 +208,7 @@ int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
 		 * this byte range must be able to fit inside
 		 * a single leaf so it will also fit inside a u32
 		 */
-		diff = offset - item_start_offset;
+		diff = disk_bytenr - item_start_offset;
 		diff = diff / root->sectorsize;
 		diff = diff * csum_size;
 
@@ -216,7 +216,11 @@ int btrfs_lookup_bio_sums(struct btrfs_root *root, struct inode *inode,
 				   ((unsigned long)item) + diff,
 				   csum_size);
 found:
-		set_state_private(io_tree, offset, sum);
+		if (dst)
+			*dst++ = sum;
+		else
+			set_state_private(io_tree, offset, sum);
+		disk_bytenr += bvec->bv_len;
 		bio_index++;
 		bvec++;
 	}
@@ -224,75 +228,8 @@ found:
 	return 0;
 }
 
-int btrfs_csum_file_bytes(struct btrfs_root *root, struct inode *inode,
-			  u64 start, unsigned long len)
-{
-	struct btrfs_ordered_sum *sums;
-	struct btrfs_sector_sum *sector_sum;
-	struct btrfs_ordered_extent *ordered;
-	char *data;
-	struct page *page;
-	unsigned long total_bytes = 0;
-	unsigned long this_sum_bytes = 0;
-
-	sums = kzalloc(btrfs_ordered_sum_size(root, len), GFP_NOFS);
-	if (!sums)
-		return -ENOMEM;
-
-	sector_sum = sums->sums;
-	sums->file_offset = start;
-	sums->len = len;
-	INIT_LIST_HEAD(&sums->list);
-	ordered = btrfs_lookup_ordered_extent(inode, sums->file_offset);
-	BUG_ON(!ordered);
-
-	while(len > 0) {
-		if (start >= ordered->file_offset + ordered->len ||
-		    start < ordered->file_offset) {
-			sums->len = this_sum_bytes;
-			this_sum_bytes = 0;
-			btrfs_add_ordered_sum(inode, ordered, sums);
-			btrfs_put_ordered_extent(ordered);
-
-			sums = kzalloc(btrfs_ordered_sum_size(root, len),
-				       GFP_NOFS);
-			BUG_ON(!sums);
-			sector_sum = sums->sums;
-			sums->len = len;
-			sums->file_offset = start;
-			ordered = btrfs_lookup_ordered_extent(inode,
-						      sums->file_offset);
-			BUG_ON(!ordered);
-		}
-
-		page = find_get_page(inode->i_mapping,
-				     start >> PAGE_CACHE_SHIFT);
-
-		data = kmap_atomic(page, KM_USER0);
-		sector_sum->sum = ~(u32)0;
-		sector_sum->sum = btrfs_csum_data(root, data, sector_sum->sum,
-						  PAGE_CACHE_SIZE);
-		kunmap_atomic(data, KM_USER0);
-		btrfs_csum_final(sector_sum->sum,
-				 (char *)&sector_sum->sum);
-		sector_sum->offset = page_offset(page);
-		page_cache_release(page);
-
-		sector_sum++;
-		total_bytes += PAGE_CACHE_SIZE;
-		this_sum_bytes += PAGE_CACHE_SIZE;
-		start += PAGE_CACHE_SIZE;
-
-		WARN_ON(len < PAGE_CACHE_SIZE);
-		len -= PAGE_CACHE_SIZE;
-	}
-	btrfs_add_ordered_sum(inode, ordered, sums);
-	btrfs_put_ordered_extent(ordered);
-	return 0;
-}
-
 int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
-		       struct bio *bio)
+		       struct bio *bio, u64 file_start, int contig)
 {
 	struct btrfs_ordered_sum *sums;
 	struct btrfs_sector_sum *sector_sum;
@@ -303,6 +240,7 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
 	unsigned long total_bytes = 0;
 	unsigned long this_sum_bytes = 0;
 	u64 offset;
+	u64 disk_bytenr;
 
 	WARN_ON(bio->bi_vcnt <= 0);
 	sums = kzalloc(btrfs_ordered_sum_size(root, bio->bi_size), GFP_NOFS);
@@ -310,16 +248,25 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
 		return -ENOMEM;
 
 	sector_sum = sums->sums;
-	sums->file_offset = page_offset(bvec->bv_page) + bvec->bv_offset;
+	disk_bytenr = (u64)bio->bi_sector << 9;
 	sums->len = bio->bi_size;
 	INIT_LIST_HEAD(&sums->list);
-	ordered = btrfs_lookup_ordered_extent(inode, sums->file_offset);
+
+	if (contig)
+		offset = file_start;
+	else
+		offset = page_offset(bvec->bv_page) + bvec->bv_offset;
+
+	ordered = btrfs_lookup_ordered_extent(inode, offset);
 	BUG_ON(!ordered);
+	sums->bytenr = ordered->start;
 
 	while(bio_index < bio->bi_vcnt) {
-		offset = page_offset(bvec->bv_page) + bvec->bv_offset;
-		if (offset >= ordered->file_offset + ordered->len ||
-		    offset < ordered->file_offset) {
+		if (!contig)
+			offset = page_offset(bvec->bv_page) + bvec->bv_offset;
+
+		if (!contig && (offset >= ordered->file_offset + ordered->len ||
+		    offset < ordered->file_offset)) {
 			unsigned long bytes_left;
 			sums->len = this_sum_bytes;
 			this_sum_bytes = 0;
@@ -333,10 +280,9 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
 			BUG_ON(!sums);
 			sector_sum = sums->sums;
 			sums->len = bytes_left;
-			sums->file_offset = offset;
-			ordered = btrfs_lookup_ordered_extent(inode,
-						      sums->file_offset);
+			ordered = btrfs_lookup_ordered_extent(inode, offset);
 			BUG_ON(!ordered);
+			sums->bytenr = ordered->start;
 		}
 
 		data = kmap_atomic(bvec->bv_page, KM_USER0);
@@ -348,13 +294,14 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
 		kunmap_atomic(data, KM_USER0);
 		btrfs_csum_final(sector_sum->sum,
 				 (char *)&sector_sum->sum);
-		sector_sum->offset = page_offset(bvec->bv_page) +
-			bvec->bv_offset;
+		sector_sum->bytenr = disk_bytenr;
 
 		sector_sum++;
 		bio_index++;
 		total_bytes += bvec->bv_len;
 		this_sum_bytes += bvec->bv_len;
+		disk_bytenr += bvec->bv_len;
+		offset += bvec->bv_len;
 		bvec++;
 	}
 	this_sum_bytes = 0;
@@ -364,11 +311,10 @@ int btrfs_csum_one_bio(struct btrfs_root *root, struct inode *inode,
 }
 
 int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
-			   struct btrfs_root *root, struct inode *inode,
+			   struct btrfs_root *root,
 			   struct btrfs_ordered_sum *sums)
 {
-	u64 objectid = inode->i_ino;
-	u64 offset;
+	u64 bytenr;
 	int ret;
 	struct btrfs_key file_key;
 	struct btrfs_key found_key;
@@ -396,13 +342,12 @@ int btrfs_csum_file_blocks(struct btrfs_trans_handle *trans,
 again:
 	next_offset = (u64)-1;
 	found_next = 0;
-	offset = sector_sum->offset;
-	file_key.objectid = objectid;
-	file_key.offset = offset;
-	btrfs_set_key_type(&file_key, BTRFS_CSUM_ITEM_KEY);
+	file_key.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
+	file_key.offset = sector_sum->bytenr;
+	bytenr = sector_sum->bytenr;
+	btrfs_set_key_type(&file_key, BTRFS_EXTENT_CSUM_KEY);
 
-	mutex_lock(&BTRFS_I(inode)->csum_mutex);
-	item = btrfs_lookup_csum(trans, root, path, objectid, offset, 1);
+	item = btrfs_lookup_csum(trans, root, path, sector_sum->bytenr, 1);
 	if (!IS_ERR(item)) {
 		leaf = path->nodes[0];
 		ret = 0;
@@ -432,8 +377,8 @@ again:
 			slot = 0;
 		}
 		btrfs_item_key_to_cpu(path->nodes[0], &found_key, slot);
-		if (found_key.objectid != objectid ||
-		    found_key.type != BTRFS_CSUM_ITEM_KEY) {
+		if (found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID ||
+		    found_key.type != BTRFS_EXTENT_CSUM_KEY) {
 			found_next = 1;
 			goto insert;
 		}
@@ -460,10 +405,10 @@ again:
 	path->slots[0]--;
 	leaf = path->nodes[0];
 	btrfs_item_key_to_cpu(leaf, &found_key, path->slots[0]);
-	csum_offset = (offset - found_key.offset) >>
+	csum_offset = (bytenr - found_key.offset) >>
 			root->fs_info->sb->s_blocksize_bits;
-	if (btrfs_key_type(&found_key) != BTRFS_CSUM_ITEM_KEY ||
-	    found_key.objectid != objectid ||
+	if (btrfs_key_type(&found_key) != BTRFS_EXTENT_CSUM_KEY ||
+	    found_key.objectid != BTRFS_EXTENT_CSUM_OBJECTID ||
 	    csum_offset >= MAX_CSUM_ITEMS(root, csum_size)) {
 		goto insert;
 	}
@@ -482,8 +427,18 @@ insert:
 	btrfs_release_path(root, path);
 	csum_offset = 0;
 	if (found_next) {
-		u64 tmp = min((u64)i_size_read(inode), next_offset);
-		tmp -= offset & ~((u64)root->sectorsize -1);
+		u64 tmp = total_bytes + root->sectorsize;
+		u64 next_sector = sector_sum->bytenr;
+		struct btrfs_sector_sum *next = sector_sum + 1;
+
+		while(tmp < sums->len) {
+			if (next_sector + root->sectorsize != next->bytenr)
+				break;
+			tmp += root->sectorsize;
+			next_sector = next->bytenr;
+			next++;
+		}
+		tmp = min(tmp, next_offset - file_key.offset);
 		tmp >>= root->fs_info->sb->s_blocksize_bits;
 		tmp = max((u64)1, tmp);
 		tmp = min(tmp, (u64)MAX_CSUM_ITEMS(root, csum_size));
@@ -510,7 +465,6 @@ found:
 	item_end = (struct btrfs_csum_item *)((unsigned char *)item_end +
 				      btrfs_item_size_nr(leaf, path->slots[0]));
 	eb_token = NULL;
-	mutex_unlock(&BTRFS_I(inode)->csum_mutex);
 	cond_resched();
 next_sector:
 
@@ -541,9 +495,9 @@ next_sector:
 	if (total_bytes < sums->len) {
 		item = (struct btrfs_csum_item *)((char *)item +
 						  csum_size);
-		if (item < item_end && offset + PAGE_CACHE_SIZE ==
-		    sector_sum->offset) {
-			    offset = sector_sum->offset;
+		if (item < item_end && bytenr + PAGE_CACHE_SIZE ==
+		    sector_sum->bytenr) {
+			bytenr = sector_sum->bytenr;
 			goto next_sector;
 		}
 	}
@@ -562,7 +516,6 @@ out:
 	return ret;
 
 fail_unlock:
-	mutex_unlock(&BTRFS_I(inode)->csum_mutex);
 	goto out;
 }